Hospital-Based System Combines AI and Genomic Surveillance to Quickly Detect Infectious Disease Outbreaks

These methods have remained largely unchanged for over a century. When two or more patients in a hospital exhibit nearly identical strains of an infection, it may be due to factors such as close proximity of hospital beds, shared medical equipment, or common healthcare providers. Typically, when clinicians notice that multiple patients have a similar infection, they notify the infection prevention team, which then reviews patient records to identify the possible source of transmission. This process is labor-intensive and often relies on busy healthcare workers identifying the shared infection between patients. While this may be feasible when patients are in the same unit, it becomes much harder to detect when patients are in different units with separate healthcare teams, and the only common factor is a visit to a procedure room. In such cases, the likelihood of detecting an outbreak before it spreads to other patients is significantly reduced. Now, a two-year trial has demonstrated that a new infectious disease detection platform can prevent outbreaks, save lives, and reduce costs for hospitals. The findings, published in Clinical Infectious Diseases, support the adoption of this system in U.S. hospitals and the creation of a national early outbreak detection database.

Researchers at the University of Pittsburgh School of Medicine (UPMC, Pittsburgh, PA, USA) have developed the Enhanced Detection System for Healthcare-Associated Transmission (EDS-HAT), which leverages increasingly affordable genomic sequencing to analyze infectious disease samples from patients. When the sequencing detects that two or more patients have nearly identical strains of an infection, it alerts the hospital’s infection prevention team, enabling them to identify the common source and halt the transmission. Without genomic sequencing, infection preventionists have no way of knowing if two patients are coincidentally infected with the same strain or if one patient was infected by the other. This lack of insight can result in infections spreading unnoticed when patients with similar infections have no obvious connection, such as being in the same unit, which allows outbreaks to grow before being detected.

On the other hand, infection preventionists may waste time and resources trying to prevent an outbreak that doesn’t exist when patients with the same infection have been exposed to unrelated sources. The study, conducted from November 2021 to October 2023 at UPMC Presbyterian Hospital, revealed that EDS-HAT prevented 62 infections and five deaths. It also resulted in savings of nearly USD 700,000 in infection treatment costs, providing a 3.2-fold return on investment. If healthcare facilities across the U.S. adopt EDS-HAT, a nationwide outbreak detection system could be established, similar to PulseNet, the U.S. Centers for Disease Control and Prevention’s network for identifying multistate outbreaks of foodborne illnesses. The researchers previously concluded that had such a system been in place, the 2023 outbreak of deadly bacteria linked to contaminated eye drops could have been detected and stopped much sooner.

“We saved lives while saving money. This isn’t theoretical – this happened in a real hospital with real patients,” said lead author Alexander Sundermann, Dr.P.H., assistant professor of infectious diseases in Pitt’s School of Medicine. “And it could easily be scaled. The more hospitals implement this practice, the more everyone benefits, not just by stopping previously undetected outbreaks within the walls of the hospital, but by finding medical device or medication-linked outbreaks sweeping the nation.”

Related Links:
UPMC